--- title: AI Agents Have the Keys to Your Computer. I Spent 8... description: AEGIS is an open-source security app that monitors and controls local AI agents, described by its creator as 'antivirus for AI agents'. It can deny a... canonical: https://daily.dev/posts/ai-agents-have-the-keys-to-your-computer-i-spent-8-months-building-an-antivirus-for-them--93ucqlipf twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: AI Agents Have the Keys to Your Computer. I Spent 8 Months Building an Antivirus for Them. | daily.dev og:description: AEGIS is an open-source security app that monitors and controls local AI agents, described by its creator as 'antivirus for AI agents'. It can deny a... og:url: https://daily.dev/posts/ai-agents-have-the-keys-to-your-computer-i-spent-8-months-building-an-antivirus-for-them--93ucqlipf og:image: https://api.daily.dev/og/posts/93UcqlIPf.png og:image:alt: AI Agents Have the Keys to Your Computer. I Spent 8 Months Building an Antivirus for Them. og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # AI Agents Have the Keys to Your Computer. I Spent 8 Months Building an Antivirus for Them. **[DEV](https://daily.dev/sources/devto)** · 13 min read · 0 upvotes · 1 comments ## Summary AEGIS is an open-source security app that monitors and controls local AI agents, described by its creator as 'antivirus for AI agents'. It can deny a configured command before it starts, require approval for an exact MCP tool call and arguments, reject known secrets on MCP routes, review skills before use, and pause or stop an agent process. The Windows beta is version 0.20.1, released under the MIT license, with macOS and Linux source paths still experimental. It also supports Cedar and supported JSON policy formats, Claude and Gemini hook integrations, imports of Cisco JSON and SARIF reports, and exports activity as JSON or CSV, an HTML session report, and audit records with SHA-256 linkage for local chain verification. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How can I stop an AI coding agent from running a dangerous command before it executes? A security tool called AEGIS puts a policy check directly in the execution path, evaluating the executable, working directory, arguments and environment before a routed command is allowed to launch. A denied decision stops the command before process creation even begins, rather than only logging it afterward. Policies can be written in Cedar or JSON formats. _Developers locking down agent permissions can track approaches like this for AI coding workflows on daily.dev._ ### How do I prevent an MCP tool call from being approved for one action and then silently reused for a different one? AEGIS's MCP gateway binds approval to the exact tool call and its exact arguments, consuming a one-time-use grant before forwarding the request. If the arguments change, a fresh approval is required, so permission for writing to one file cannot carry over to writing a different file like .env. The gateway supports stdio, loopback HTTP and pinned HTTPS routes. _Teams wiring up MCP tool access can compare approval models like this through daily.dev before trusting an agent._ ### What license is the AEGIS AI agent security tool released under and what platforms does it support? AEGIS is released under the MIT license, with a downloadable Windows beta at version 0.20.1 as the public delivery. macOS and Linux source paths remain experimental rather than officially supported. The tool includes a validating MCP gateway, command execution policies, skill review, and audit export features including JSON, CSV, and SHA-256-linked daily records. _Developers evaluating open-source agent security tooling can follow releases like this one via daily.dev._ ## Community discussion Top comments from developers on daily.dev. **@bickov** · 0 upvotes > How does the known secrets check treat images? Playwright MCP's browser_take_screenshot returns a PNG, and a key visible in a terminal or in a .env file open in an editor goes through as pixels, so a text pattern match never sees it. Is OCR on image tool results planned, or out of scope for the 0.20.1 Windows beta? ## Similar posts on daily.dev - [Introducing the AI Agent Security Scanner for IDEs: Verify Your Agents](https://daily.dev/posts/introducing-the-ai-agent-security-scanner-for-ides-verify-your-agents-lk73jmoc4) · Cisco · 1 upvotes · 0 comments - [I built a fully local agentic AI setup with one open-source app, and it never touches the cloud](https://daily.dev/posts/i-built-a-fully-local-agentic-ai-setup-with-one-open-source-app-and-it-never-touches-the-cloud-xe4auoith) · XDA Developers · 0 upvotes · 0 comments - [How Kaspersky solutions protect against AI threats](https://daily.dev/posts/how-kaspersky-solutions-protect-against-ai-threats-dgaah8i4d) · Securelist · 1 upvotes · 0 comments - [Software developers are becoming agent builders](https://daily.dev/posts/software-developers-are-becoming-agent-builders-qvdhvvauw) · LeadDev · 0 upvotes · 0 comments --- Tags: [#ai](https://daily.dev/tags/ai), [#security](https://daily.dev/tags/security), [#open-source](https://daily.dev/tags/open-source), [#ai-agents](https://daily.dev/tags/ai-agents), [#mcp](https://daily.dev/tags/mcp), [#claude-code](https://daily.dev/tags/claude-code) [View this post on daily.dev](https://daily.dev/posts/ai-agents-have-the-keys-to-your-computer-i-spent-8-months-building-an-antivirus-for-them--93ucqlipf) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"AI Agents Have the Keys to Your Computer. I Spent 8 Months Building an Antivirus for Them.","url":"https://daily.dev/posts/ai-agents-have-the-keys-to-your-computer-i-spent-8-months-building-an-antivirus-for-them--93ucqlipf","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/ai-agents-have-the-keys-to-your-computer-i-spent-8-months-building-an-antivirus-for-them--93ucqlipf"},"datePublished":"2026-10-27T11:19:24.908Z","dateModified":"2026-10-27T11:19:53.659Z","description":"AEGIS is an open-source security app that monitors and controls local AI agents, described by its creator as 'antivirus for AI agents'. It can deny a...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ceec7a739390d98de7617567dad44005?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ceec7a739390d98de7617567dad44005?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"DEV","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"DEV","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/devto","url":"https://daily.dev/sources/devto"},"commentCount":1,"discussionUrl":"https://daily.dev/posts/ai-agents-have-the-keys-to-your-computer-i-spent-8-months-building-an-antivirus-for-them--93ucqlipf","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":1}],"keywords":"ai,security,open-source,ai-agents,mcp,claude-code","timeRequired":"PT13M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"DEV","item":"https://daily.dev/sources/devto"},{"@type":"ListItem","position":3,"name":"AI Agents Have the Keys to Your Computer. I Spent 8 Months Building an Antivirus for Them."}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/ai-agents-have-the-keys-to-your-computer-i-spent-8-months-building-an-antivirus-for-them--93ucqlipf","comment":[{"@type":"Comment","text":"How does the known secrets check treat images? Playwright MCP’s browser_take_screenshot returns a PNG, and a key visible in a terminal or in a .env file open in an editor goes through as pixels, so a text pattern match never sees it. Is OCR on image tool results planned, or out of scope for the 0.20.1 Windows beta?","datePublished":"2026-10-28T08:22:34.697Z","url":"https://daily.dev/posts/93UcqlIPf#c-Gusidnte4","author":{"@type":"Person","name":"ab","url":"https://daily.dev/bickov","image":"https://lh3.googleusercontent.com/a/ACg8ocKwOmJ31UWqM4ZXB5ms-yRdFR4BYzyb7FPBRpg1O6DV0MIAF2EP=s96-c"}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/ai-agents-have-the-keys-to-your-computer-i-spent-8-months-building-an-antivirus-for-them--93ucqlipf#faq","mainEntity":[{"@type":"Question","name":"How can I stop an AI coding agent from running a dangerous command before it executes?","acceptedAnswer":{"@type":"Answer","text":"A security tool called AEGIS puts a policy check directly in the execution path, evaluating the executable, working directory, arguments and environment before a routed command is allowed to launch. A denied decision stops the command before process creation even begins, rather than only logging it afterward. Policies can be written in Cedar or JSON formats. Developers locking down agent permissions can track approaches like this for AI coding workflows on daily.dev."}},{"@type":"Question","name":"How do I prevent an MCP tool call from being approved for one action and then silently reused for a different one?","acceptedAnswer":{"@type":"Answer","text":"AEGIS's MCP gateway binds approval to the exact tool call and its exact arguments, consuming a one-time-use grant before forwarding the request. If the arguments change, a fresh approval is required, so permission for writing to one file cannot carry over to writing a different file like .env. The gateway supports stdio, loopback HTTP and pinned HTTPS routes. Teams wiring up MCP tool access can compare approval models like this through daily.dev before trusting an agent."}},{"@type":"Question","name":"What license is the AEGIS AI agent security tool released under and what platforms does it support?","acceptedAnswer":{"@type":"Answer","text":"AEGIS is released under the MIT license, with a downloadable Windows beta at version 0.20.1 as the public delivery. macOS and Linux source paths remain experimental rather than officially supported. The tool includes a validating MCP gateway, command execution policies, skill review, and audit export features including JSON, CSV, and SHA-256-linked daily records. Developers evaluating open-source agent security tooling can follow releases like this one via daily.dev."}}]} ```